Process Discipline as the Key Variable in Ai-Assisted Enterprise Software Development: a Natural Experiment

Findings: Same team, same AI tools, different process conditions: 18/18 enterprise dimensions satisfied at 8-10x elite productivity benchmarks under a structured SDLC, versus 2/18 dimensions with no tests, no quality pipelines, and no code review when process authority shifted to non-technical stakeholders. Measured across 287 FTE-days, 1.48 million lines added, 811,000 lines removed, and 18 enterprise dimensions derived from SOC 2, NIST, OWASP, DORA, CIS, and CNCF frameworks. Scale: No comparable study in the AI-assisted development literature approaches this duration or granularity. Peng et al. (2023) measured a single isolated task. METR (2025) tested 16 developers on individual issues. DeputyDev (2025) observed 300 engineers but had no unstructured comparison arm. This study spans 13.7 FTE-months of sustained enterprise development with commit-level traceability across multiple codebases. Literature gaps addressed: (1) No published study isolates process discipline as a controlled variable in AI-assisted development. This paper presents a natural experiment holding team and tools constant while varying the development process through an exogenous organizational change. (2) The speed literature produces contradictory findings (55.8% speedups vs. 19% slowdowns) with no reconciliation. This paper argues these are measurements of different process conditions, not conflicting results. (3) No published benchmark exists for sustained AI-assisted commit rates; this paper reports 10.7 commits/FTE-day over 287 FTE-days. (4) DORA's "amplifier" thesis rests on correlational survey data; this paper provides project-level evidence with a causal mechanism. (5) Model collapse research (Shumailov et al. 2024, Nature) has not been connected to practical codebase quality; this paper identifies the clean starting codebase as a multiplicative requirement grounded in generation loss theory. (6) The role of organizational governance in AI-assisted development quality has not been empirically demonstrated; this paper documents a quality collapse caused by an organizational decision, not a technical one. All metrics derived from git history, source code analysis, and published industry benchmarks. Methodology described for replication.